Thread: [mpls-linux-general] Diffserv & Van den Berghe Patch

Status: Beta

Brought to you by: jleu

mpls-linux-general

[mpls-linux-general] Diffserv & Van den Berghe Patch

From: Filip R. <fi...@cs...> - 2001-05-17 12:03:08

Hi,

I've just updated my mpls configuration to mpls-linux-0.990 on kernel
2.4.2 patched with Van Den Berghe's tc patch mpls-tc.2.4.1-0502.

Questions:
1. When I try to configure a LER with two outgoing label-entries with
the very same label but different EXP values, the second one doesn't get
registered. Aren't I supposed to have one label/EXP entry for each
aggregated flow that I wish to support at the LER? Whats wrong?

mplsadm -v -A -O gen:120/1:eth0:ipv4:10.1.1.2
mplsadm -v -A -O gen:120/2:eth0:ipv4:10.1.1.2


2. Same holds when I want to configure the LSR to do label swapping
based on label/EXP pairs. The second one doesn't get registered, is it
supposed to be like this?

mplsadm -v -A -I gen:113/4:0 -O gen:122/2:eth0:ipv4:10.1.1.2 -B
mplsadm -v -A -I gen:113/3:0 -O gen:122/3:eth0:ipv4:10.1.1.2 -B


Thankful for any help, hints or pointers!

/Filip Rosenbaum

Re: [mpls-linux-general] Diffserv & Van den Berghe Patch

From: James R. L. <jl...@mi...> - 2001-05-17 14:14:24

I'm not going to be very usefull in debugging this.  Maybe Steven could
help out?

Jim

On Thu, May 17, 2001 at 10:01:39PM +1000, Filip Rosenbaum wrote:
> Hi,
> 
> I've just updated my mpls configuration to mpls-linux-0.990 on kernel
> 2.4.2 patched with Van Den Berghe's tc patch mpls-tc.2.4.1-0502.
> 
> Questions:
> 1. When I try to configure a LER with two outgoing label-entries with
> the very same label but different EXP values, the second one doesn't get
> registered. Aren't I supposed to have one label/EXP entry for each
> aggregated flow that I wish to support at the LER? Whats wrong?
> 
> mplsadm -v -A -O gen:120/1:eth0:ipv4:10.1.1.2
> mplsadm -v -A -O gen:120/2:eth0:ipv4:10.1.1.2
> 
> 
> 2. Same holds when I want to configure the LSR to do label swapping
> based on label/EXP pairs. The second one doesn't get registered, is it
> supposed to be like this?
> 
> mplsadm -v -A -I gen:113/4:0 -O gen:122/2:eth0:ipv4:10.1.1.2 -B
> mplsadm -v -A -I gen:113/3:0 -O gen:122/3:eth0:ipv4:10.1.1.2 -B
> 
> 
> Thankful for any help, hints or pointers!
> 
> /Filip Rosenbaum
> 
> 
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general

-- 
James R. Leu

Re: [mpls-linux-general] Diffserv & Van den Berghe Patch

From: Steven V. d. B. <ste...@in...> - 2001-05-17 15:21:04

Hi Filip (and all MPLSers),

First of all, my excuses for the late response and the follow-up on the
mpls patches. It's been quite busy lately, and although the patches are
almost stable now, i haven't put them through the final test phase. This
will be done asap. These will hopefully resolve all the remaining problems
(i'll release 3 patches: the DS patch, multiple table support and
statistics support).

Concerning your problem:
  - i think both labels/exp combinations get registered but don't get
printed (this is just from the top of my hat).
  - if this isn't so, i'm very sorry but the patch used was indeed a
dreadful thing with lots of bugs in it. I'll check if it works with the new
patches and release them asap.

To be certain, could you check with:

mplsadm -v -A -B -O gen:120/1:eth0:ipv4:10.1.1.2 -f <some fec>
mplsadm -v -A -B -O gen:120/2:eth0:ipv4:10.1.1.2 -f <another fec>

that there is indeed no possibility to set multiple EXPs (PS use ethereal
to check which EXP the packets are getting)

Thanks,
Steven

Filip Rosenbaum wrote:

> Hi,
>
> I've just updated my mpls configuration to mpls-linux-0.990 on kernel
> 2.4.2 patched with Van Den Berghe's tc patch mpls-tc.2.4.1-0502.
>
> Questions:
> 1. When I try to configure a LER with two outgoing label-entries with
> the very same label but different EXP values, the second one doesn't get
> registered. Aren't I supposed to have one label/EXP entry for each
> aggregated flow that I wish to support at the LER? Whats wrong?
>
> mplsadm -v -A -O gen:120/1:eth0:ipv4:10.1.1.2
> mplsadm -v -A -O gen:120/2:eth0:ipv4:10.1.1.2
>
> 2. Same holds when I want to configure the LSR to do label swapping
> based on label/EXP pairs. The second one doesn't get registered, is it
> supposed to be like this?
>
> mplsadm -v -A -I gen:113/4:0 -O gen:122/2:eth0:ipv4:10.1.1.2 -B
> mplsadm -v -A -I gen:113/3:0 -O gen:122/3:eth0:ipv4:10.1.1.2 -B
>
> Thankful for any help, hints or pointers!
>
> /Filip Rosenbaum
>
> _______________________________________________
> mpls-linux-general mailing list
> mpl...@li...
> http://lists.sourceforge.net/lists/listinfo/mpls-linux-general

--
Steven Van den Berghe
ste...@in...
Workgroup Broadband-Networks
Department Information Technology
Ghent University - Belgium
Phone:  +32 (0)9 267 35 86 | Fax  :  +32 (0)9 267 35 99
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
Unix is user friendly - it's just picky about it's
friends.


*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

Re: [mpls-linux-general] Diffserv & Van den Berghe Patch

From: Filip R. <fi...@cs...> - 2001-05-18 06:57:14

Thanks for your quick answer,

I checked the actual EXP field with ethereal as suggested, and only the first
label/EXP will be used. In fact, the second FEC will be mapped to the first
label/EXP.

Question:
Is there any way to base the label-selection at the ingress LER on anything
else than the so called FEC which appears to be just an masked IP-destination
address in the linux-mpls?

It is a bit limiting not to be able to distinguish different flows going to
the same destination network/host from each other. Will the coming patch allow
tc to do both the classification and label-selection/assigning?


Regards,
Filip Rosenbuam



Steven Van den Berghe wrote:

> Hi Filip (and all MPLSers),
>
> First of all, my excuses for the late response and the follow-up on the
> mpls patches. It's been quite busy lately, and although the patches are
> almost stable now, i haven't put them through the final test phase. This
> will be done asap. These will hopefully resolve all the remaining problems
> (i'll release 3 patches: the DS patch, multiple table support and
> statistics support).
>
> Concerning your problem:
>   - i think both labels/exp combinations get registered but don't get
> printed (this is just from the top of my hat).
>   - if this isn't so, i'm very sorry but the patch used was indeed a
> dreadful thing with lots of bugs in it. I'll check if it works with the new
> patches and release them asap.
>
> To be certain, could you check with:
>
> mplsadm -v -A -B -O gen:120/1:eth0:ipv4:10.1.1.2 -f <some fec>
> mplsadm -v -A -B -O gen:120/2:eth0:ipv4:10.1.1.2 -f <another fec>
>
> that there is indeed no possibility to set multiple EXPs (PS use ethereal
> to check which EXP the packets are getting)
>
> Thanks,
> Steven
>
> Filip Rosenbaum wrote:
>
> > Hi,
> >
> > I've just updated my mpls configuration to mpls-linux-0.990 on kernel
> > 2.4.2 patched with Van Den Berghe's tc patch mpls-tc.2.4.1-0502.
> >
> > Questions:
> > 1. When I try to configure a LER with two outgoing label-entries with
> > the very same label but different EXP values, the second one doesn't get
> > registered. Aren't I supposed to have one label/EXP entry for each
> > aggregated flow that I wish to support at the LER? Whats wrong?
> >
> > mplsadm -v -A -O gen:120/1:eth0:ipv4:10.1.1.2
> > mplsadm -v -A -O gen:120/2:eth0:ipv4:10.1.1.2
> >
> > 2. Same holds when I want to configure the LSR to do label swapping
> > based on label/EXP pairs. The second one doesn't get registered, is it
> > supposed to be like this?
> >
> > mplsadm -v -A -I gen:113/4:0 -O gen:122/2:eth0:ipv4:10.1.1.2 -B
> > mplsadm -v -A -I gen:113/3:0 -O gen:122/3:eth0:ipv4:10.1.1.2 -B
> >
> > Thankful for any help, hints or pointers!
> >
> > /Filip Rosenbaum
> >
> > _______________________________________________
> > mpls-linux-general mailing list
> > mpl...@li...
> > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general
>
> --
> Steven Van den Berghe
> ste...@in...
> Workgroup Broadband-Networks
> Department Information Technology
> Ghent University - Belgium
> Phone:  +32 (0)9 267 35 86 | Fax  :  +32 (0)9 267 35 99
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> Unix is user friendly - it's just picky about it's
> friends.
>
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

Re: [mpls-linux-general] Diffserv & Van den Berghe Patch

From: Steven V. d. B. <ste...@in...> - 2001-05-18 07:35:39

Hi Filip,

Please see my comments inline (i'm on some time pressure here, so they're really
short),

Filip Rosenbaum wrote:

> Thanks for your quick answer,
>

No problemo

>
> I checked the actual EXP field with ethereal as suggested, and only the first
> label/EXP will be used. In fact, the second FEC will be mapped to the first
> label/EXP.
>

Oops, another bug on the wall

>
> Question:
> Is there any way to base the label-selection at the ingress LER on anything
> else than the so called FEC which appears to be just an masked IP-destination
> address in the linux-mpls?
>

The "multiple table patch" can solve this. Here you can set up different routing
tables for different sources, firewalling results, etc.
I mostly use iptables for setting up firewalling rules (that do the
classification on whatever you want) and iproute2 for more the routing table
configs.
You can find more info on

http://snafu.freedom.org/linux2.2/iproute-notes.html
http://netfilter.samba.org/

So, i hope this helps a little,

Best regards,
Steven

--
Steven Van den Berghe
ste...@in...
Workgroup Broadband-Networks
Department Information Technology
Ghent University - Belgium
Phone:  +32 (0)9 267 35 86 | Fax  :  +32 (0)9 267 35 99
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
If your computer says: Printer out of Paper, this problem
cannot be resolved by continuously clicking the 'OK'
button.

*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*